The Quiet Collapse in the Middle Overs: A 'Recovery Efficiency Index' Investigation into Bangladesh's T20 Batting

The Quiet Collapse in the Middle Overs

The Silence of a Single Over

Zahur Ahmed Chowdhury Stadium, Chattogram, March 15, 2026. Bangladesh were 62/2 after eight overs — the innings was on track. Then came 34 runs and four wickets across the next seven overs. At the fifteen-over mark the score read 96/6. The innings finished at 148/9.

That evening I measured the ball-by-ball trace rather than the scorecard. Between overs eight and fifteen, the dot-ball rate jumped from 31 percent to 52 percent; the boundary rate slid from 14.6 percent to 6.2 percent. The collapse did not happen in one wicket — it was a quiet, systemic drift, ball after ball. The scorecard never records it, because a scorecard counts runs, not the interval between them.

Since launching "Expected Truth" from Khulna in 2026, I have kept one habit: write the definition before writing the verdict. This piece is the product of that habit. The numbers did not break the model; they exposed where the model was blind.

Context: Why the Middle Overs

A T20 match is usually decided in two places — the fielding-restriction overs and the death-overs panic. Overs seven to fifteen are treated as the zone of control: runs come slowly, but a wicket there flips the game. For Bangladesh this window is the most deceptive one. Fewer boundaries fall, fewer runs arrive, so the drama appears smaller. Nobody asks at the end of a match what happened between the ninth and thirteenth overs.

Bangladesh's batting identity was built on a slow, low-bounce Mirpur–Sylhet–Chattogram circuit, where rotation rather than stroke-play does the work, and the foundation of rotation is survival. Building an index of that survival revealed something: the problem is not the run rate, it is the structure of continuity.

Methodology Note

My sample: 26 T20Is between January 1, 2026 and December 31, 2026 — 24 completed, 2 rain-affected. That gives 216 middle-over overs, or 1,234 balls. Every ball was tagged across four dimensions: phase (powerplay/middle/death), bowler type (spin/pace), post-wicket state, and venue baseline.

Three indices were defined, and written down before the investigation began, so the definitions could not be bent later to fit the data:

• Pressure Over (PO): any over in which a team's run rate falls 20 percent or more below its own phase baseline.
• Recovery Efficiency Index (REI): runs scored in the twelve balls after a wicket, divided by the phase-and-venue expected runs. 1.00 is baseline; below 1.00 is post-collapse contraction.
• Phase Leverage (PL): a weighted product of wickets in hand, balls remaining, and the rising required rate.

Baselines are the two-year averages of the top eight T20 sides, plus a separate South Asian venue-specific baseline, because the character of overs seven to fifteen in India is not comparable to Mirpur.

Core Analysis: The Chain of Evidence

Phase Splits

Bangladesh over the two-year window:

• Powerplay (1–6): run rate 7.8 | dots 38.4% | boundaries 15.2%
• Middle overs (7–15): run rate 6.9 | dots 44.1% | boundaries 9.8%
• Death overs (16–20): run rate 8.9 | dots 31.2% | boundaries 14.1%

The top-eight middle-over baseline: run rate 8.2 | dots 36.8% | boundaries 12.4%. The deficit is 1.3 in run rate, but 2.6 points in boundary rate and 7.3 points in dot balls. The problem is not aggression; the problem is rotation. Aggression returns at the death. Rotation is lost in the middle.

The Architecture of Dot Balls

Within the middle overs, 27.4 percent of Bangladesh's dot balls fall on the fifth to seventh delivery of the over. An over that begins with a single ends up stuck. Bangladesh take a single off the first two balls of a middle over 54 percent of the time, yet the six-ball score sometimes stalls at 2 or 3. That picture is not one batter's failure; it is a failure in the weighting of the batting order.

Spin Versus Pace

Strike rates in the middle overs against Bangladesh:

• Versus spin: 108.4 | dots 46.2% | boundaries 8.1%
• Versus pace: 128.7 | dots 39.4% | boundaries 11.6%

Over the last two years opposition captains have bowled spin for 49.6 percent of Bangladesh's middle-over deliveries, and spin has produced 63 percent of the dot balls. Opposition teams are not guessing; they have read the pitch. Bangladesh's problem is not the pitch either — it is a decline in sweep and late-cut usage. Across the six venues we call home, the rate of scoring outside the square boundary is less than half the rate inside it.

Wicket Clustering

This is the most uncomfortable finding. The probability of a second wicket falling within twelve balls of a first:

• Bangladesh: 34.2%
• Top-eight baseline: 26.4%
• South Asian venue baseline: 29.1%

A gap of more than eight points is hard to dismiss as luck. Against the venue baseline the gap narrows to 5.1 points but survives. Part of the clustering is simply the settling-in period of a new batter — what I call settling lag. For Bangladesh that lag averages 9.4 balls; for the top eight it is 6.8.

The Recovery Efficiency Index

In the twelve balls after a wicket:

• Bangladesh scored 8.4 against an expected 9.8 — REI 0.86
• Top-eight average — REI 1.04
• Individually: India 1.14, Australia 1.09, England 1.06, Afghanistan 0.97

An REI below 1.00 means a side does not merely absorb the shock of a wicket; it carries that shock into the following overs. The index's relationship with results looks superb at first glance: in matches with REI below 0.90, Bangladesh are 3-8; with REI at or above 1.00, they are 9-2. I am suspicious of the direction of that relationship — more on that below.

Venue Split

This is where the real picture emerges. At home, Bangladesh's middle-over run rate is 6.9, while the all-team middle-over run rate at those same venues is 6.6. On Mirpur, Sylhet and Chattogram pitches, Bangladesh are actually outperforming the venue. Away, their middle-over run rate is 6.4 against a venue baseline of 7.9 — a deficit of 1.5. The hole is not a curse of home pitches; it is the direction in which Bangladesh read a pitch. Abroad, the slow-pitch rotation model has to change, and Bangladesh's batters do not change fast enough.

Chasing Versus Setting

Chasing, the middle-over run rate is 6.4; setting, 7.3. That is a confidence figure, not a capability figure. Over the two years Bangladesh used eleven different top-six combinations across 26 matches. Within that instability, a chase demands a fixed role that this batting order does not yet possess.

Batter-Level Picture

Middle-over strike rates at a minimum of 100 balls:

• Towhid Hridoy: 132.6 (218 balls)
• Mehidy Hasan Miraz: 124.1 (146 balls)
• Jaker Ali: 119.8 (112 balls)
• Litton Das: 118.3 (264 balls)
• Najmul Hossain Shanto: 111.4 (302 balls)
• Mahmudullah: 109.7 (178 balls)

Read the list and it is tempting to conclude that Hridoy is the answer. Be careful: 218 balls is a hint, not proof of an established trend, and the standard error remains wide. I don't chase outliers; I follow them until they confess — Hridoy's number has not confessed yet.

The Contrarian Angle: Correlation Is Not Causation

That striking REI-to-wins relationship collapses under my own scepticism. A team that is winning naturally bats with less pressure after a wicket — so winning raises REI, not only the reverse. To reduce that effect I ran a holdout test: built the model on the first thirteen matches, tested it on the remaining eleven. The result: predictive power fell by roughly 6.2 percentage points. The index is not useless, but it is not sufficient alone.

Second, wicket clustering may carry a selection effect — the team that attacks more loses more wickets and scores more runs. How much of Bangladesh's 34.2 percent is the product of aggressive shot selection and how much is excessive caution has not been separated yet. Version two of this model adds a coverage-zone layer.

Third, and most important: venue and dew are the blind spots of my model. In that March match in Chattogram, the dew in the second innings was heavy enough that spinners could not use a sliding grip; my REI baseline does not capture that. During the 2026 Sylhet leg, fog pushed the dot-ball rate up by six points in day-light matches, and the model still logs that as batting failure. That error is mine, not the venue's.

Takeaway: A Pre-Registered Signal for the Next Series

Stated in advance, as always: if Bangladesh's REI stays below 0.95 across the first two matches of the three-match series starting in the second week of January 2026, I put the probability of losing at least one match above 65 percent. I am also locking in the revision rule now — if the dew index in the second innings exceeds 3 out of 5, I will re-baseline upward by 0.08.

Expected truth is not a verdict; it is a pending hypothesis that has earned the right to be proven wrong by good process. After the next series we will not only check whether the result matched. We will check whether our phase definitions and wicket tagging held up against reality. Since that small Khulna newsletter in 2026, one lesson has stayed with me: a model's most valuable output is the moment it admits it was wrong.